messenger_stats.py

#!/usr/bin/env python3
import os, glob, json, pickle, sys, datetime
from shutil import rmtree


def overview():
        groups, dms = {}, {}

        for name in os.listdir('.'):
            if not os.path.isdir(name):
                continue
            files = glob.glob(f"{name}/message_*.json")
            if not files:
                rmtree(name)
            else:
                for f in files:
                    content = json.load(open(f))
                    key = name[:name.find("_")]
                    length = len(content["messages"])
                    if length < 10:
                        continue
                    if len(content["participants"]) > 2:
                        if key in groups:
                            groups[key] += length
                        else:
                            groups[key] = length
                    else:
                        if key in dms:
                            dms[key] += length
                        else:
                            dms[key] = length

        pickle.dump(groups, open('groups.pkl','wb'))
        pickle.dump(dms,open('dms.pkl','wb'))



        for pos, group in enumerate(sorted(groups, key=lambda x: -groups.get(x))):
            print(f"{pos}. {group} => {groups.get(group)}")

        print()
        for pos, dm in enumerate(sorted(dms,key=lambda x: -dms.get(x))):
            print(f"{dm} {dms.get(dm)}")


by_year, by_month, by_week, by_day = {}, {}, {}, {}

def overtime(person, base_yr):
        global by_year, by_month, by_week, by_day
        messages_dir = [x for x in glob.glob(f"{person}*") if os.path.isdir(x)]
        if not messages_dir:
                print("Cant find person")
                return
        files = glob.glob(f"{messages_dir[0]}/message_*.json")
        if not files:
                print(messages_dir)
                print("No messages")
                return

        for f in files:
                content = json.load(open(f))
                for message in content["messages"]:
                        date = datetime.datetime.fromtimestamp(message["timestamp_ms"]/1000.0)
                        year, month, week, day = date.year, date.month, date.isocalendar()[1], date.timetuple().tm_yday
                        if year > base_yr:
                                month += 12*(year-base_yr)
                                week += 52*(year-base_yr)
                                day += 365*(year-base_yr)

                        if day == 259:
                                print(message.get("content", None), message["timestamp_ms"], end='\n')

                        if year in by_year:
                                by_year[year] += 1
                        else:
                                by_year[year] = 1
                        if month in by_month:
                                by_month[month] += 1
                        else:
                                by_month[month] = 1
                        if week in by_week:
                                by_week[week] += 1
                        else:
                                by_week[week] = 1
                        if day in by_day:
                                by_day[day] += 1
                        else:
                                by_day[day] = 1

                for name, d in {'by_year': by_year, 'by_month': by_month, 'by_week': by_week, 'by_day': by_day}.items():
                        with open(f"../{person}_{name}.txt", "w") as w:
                                for k in sorted(d):
                                        w.write(f"{k},{d.get(k)}\n")


def main():
        if len(sys.argv) == 3:
                overtime(sys.argv[1], int(sys.argv[2]))
        else:
                overview()


if __name__ == '__main__':
        main()